Re: Is SMTP really part of the Web Architecture?

Elliotte Rusty Harold wrote:
> Consequently, I am suspicious that principles designed for good HTTP/ 
> FTP/etc.  may be actively harmful to good SMTP (and vice versa). I do 
> not think this document should attempt to prescribe architecture for 
> SMTP in particular or e-mail in general. It's just too different.

Just FYI, the concensus among people who are thinking about futuristic 
designs for Mail-NG is that the next SMTP should use a pull model rather 
than a push model, so that the person sending the message pays the price 
of delivery in bandwidth. Of course you need a notification mechanism to 
tell you WHEN to pull. Many of these people also agree that the pull 
part might as well use Web infrastructure (URIs and HTTP GET). 
Conversely, the Web infrastructure desperately needs a push notification 
mechanism (e.g. for cache invalidation). So I don't think that they are 
as different as you believe.

That said, disloging a deployed protocol is extremely difficult so I 
don't expect it to happen soon, if ever. But it wouldn't be a bad thing 
to unify mail and the Web at all.

  Paul Prescod

Received on Tuesday, 10 September 2002 19:04:41 UTC